aiwiki.page
English
Science / Biology / hi-c

Hi-C

Hi-C is a sequencing-based method that measures genome-wide DNA contact frequencies to investigate chromosome organization and support genome assembly.

29 keywords3 linked from9 not yet writtenWritten by AI
Molecular Biolog…GenomeDNA SequencingChromosomeChromatinDNACell NucleusProteinHi-C

Hi-C is a molecular biology technique for measuring how frequently different regions of a genome are captured in spatial proximity. It combines chemical fixation, DNA fragmentation, proximity ligation, and high-throughput DNA sequencing to generate genome-wide maps of contacts within and between chromosomes. These maps reveal features of chromatin organization, including chromosome territories, compartments, domains, and focal interactions. Hi-C measures population-level contact frequencies rather than directly photographing chromosomes or determining a unique three-dimensional structure.

Principle

A chromosome is a long DNA molecule associated with proteins and folded within a cell nucleus. Regions separated by a large distance along its sequence may lie close together in space. Hi-C captures a subset of these spatial relationships by joining DNA fragments that are physically near one another under the conditions of the experiment.

The method belongs to the family of chromosome conformation capture techniques, often called 3C-based methods. Conventional 3C measures interactions between selected genomic regions; Hi-C extends the approach to an approximately all-versus-all survey without requiring predetermined target loci.

A Hi-C sequencing read pair usually identifies the genomic positions of the two ends of a captured ligation product. After filtering, many such pairs are aggregated into a contact map. More frequent recovery of a pair of regions generally indicates a higher probability of capture in proximity, but recovery also depends on biochemical efficiency, sequence properties, and data processing.

“Contact” is therefore an operational term. It does not necessarily mean direct DNA–DNA binding, a stable molecular interaction, or a precisely known physical separation.

Experimental workflow

Protocols vary, but conventional Hi-C includes several core stages.

Fixation and fragmentation

Cells or nuclei are usually treated with formaldehyde, which creates crosslinks involving DNA and nearby proteins. This helps preserve aspects of chromatin organization during subsequent processing.

The crosslinked DNA is then fragmented, commonly using a restriction enzyme. The choice of enzyme affects the distribution and size of fragments and therefore influences which contacts can be recovered. Some related protocols use other fragmentation strategies.

End labeling and proximity ligation

In a standard restriction-based workflow, fragment ends are filled in with nucleotides that include biotin-labeled molecules. The resulting ends are joined by DNA ligase. Because nearby fragments are more likely to be joined, the ligation products retain information about their spatial association.

Early protocols performed ligation after substantial disruption of the nuclear material. In situ Hi-C instead performs proximity ligation largely within permeabilized nuclei, helping preserve local organization and reduce some sources of spurious ligation.

Library preparation and sequencing

Crosslinks are reversed, DNA is purified and fragmented further if required, and biotin-containing ligation junctions are enriched using streptavidin. Sequencing adapters are added, and the library is commonly amplified using polymerase chain reaction before paired-end sequencing.

The two sequenced ends are aligned to a reference genome. A useful pair must be distinguishable from products such as unligated fragments, self-ligated fragments, amplification duplicates, or ambiguously mapped reads.

The exact order of processing steps differs among protocols. Not every sequencing read corresponds to an informative chromatin contact.

Contact maps and computational analysis

Matrix representation

Hi-C data are commonly represented as a symmetric contact matrix. The genome is divided into intervals, or bins, and each matrix entry records the number of accepted contacts between two bins.

If CijC_{ij} is the contact count between bins ii and jj, then ideally

Cij=Cji.C_{ij}=C_{ji}.

Contact maps are often displayed as heat maps. Within a chromosome, a strong diagonal arises because regions close together along the DNA sequence are usually more likely to contact one another than distant regions. Contacts within one chromosome are called cis contacts; those between different chromosomes are called trans contacts.

The genomic bin size is not the same as the physical resolution of an image. A map displayed in small bins may still contain too few observations to support reliable interpretation at that scale.

Filtering and quality assessment

Analysis generally removes duplicate molecules, low-quality alignments, and recognizable experimental artifacts. Common quality measures include the number of unique valid contact pairs, library complexity, proportions of cis and trans contacts, and the distribution of contacts by genomic separation.

These measures must be interpreted in context. For example, an unusual trans-contact fraction may reflect sample quality, experimental artifacts, or genuine biological differences.

Reliable comparisons also depend on biological replication, compatible experimental procedures, and an appropriate reference assembly.

Normalization and distance dependence

Raw counts are influenced by restriction-site density, GC content, sequence mappability, fragment length, and other technical factors. Normalization methods attempt to reduce these biases.

A common family of approaches balances the matrix so that retained bins have comparable total contact counts. Schematically, a corrected matrix can be written as

Mij=wiCijwj,M_{ij}=w_i C_{ij}w_j,

where wiw_i and wjw_j are estimated correction weights. Such balancing assumes that systematic differences in coverage can be separated sufficiently from biological variation; it does not remove every source of bias.

Because contact frequency depends strongly on genomic separation, some analyses compare an observed count with the average count expected at a similar separation:

Rij=MijE(sij),R_{ij}=\frac{M_{ij}}{E(s_{ij})},

where sijs_{ij} is the genomic separation and E(sij)E(s_{ij}) is a distance-dependent expectation. This observed-to-expected transformation helps reveal patterns that would otherwise be obscured by the diagonal.

Comparisons between samples require particular care when total contact levels, chromosome copy numbers, or broad chromatin organization differ. Normalization can obscure genuine global changes or create apparent local differences if its assumptions are inappropriate.

Major features revealed by Hi-C

Chromosome territories

In many eukaryotic interphase nuclei, individual chromosomes preferentially occupy distinct, partly overlapping nuclear regions known as chromosome territories. Hi-C supports this organization through the predominance of contacts within chromosomes over contacts between chromosomes.

Territories are not impermeable compartments: interchromosomal contacts occur, and their patterns vary with cell type and nuclear organization.

A and B compartments

Large-scale contact maps often show a plaid or checkerboard pattern reflecting preferential association among genomic regions with similar chromatin properties. These patterns are commonly described as A and B compartments.

A-compartment regions tend to be relatively accessible and associated with active gene expression, whereas B-compartment regions tend to be less transcriptionally active and more compact. These are statistical associations, not universal classifications of every sequence in a compartment.

Compartments are often inferred using eigenvector-based analyses of transformed contact matrices. The mathematical sign of an eigenvector is arbitrary, so external information is needed to label its two directions as A and B. Higher-depth data can distinguish more detailed compartment subdivisions.

Topologically associating domains

A topologically associating domain, or TAD, is a genomic interval that shows enriched internal contacts relative to surrounding regions at a chosen scale. TADs often appear as triangular regions along the diagonal of a contact map.

Their boundaries depend partly on the detection algorithm, bin size, and experimental depth. Domains can be nested, and the term does not imply that each domain is a permanently isolated structure in every cell.

In mammalian genomes, many domain boundaries are associated with the DNA-binding protein CTCF and the cohesin complex. However, contact domains can arise through different mechanisms, and similarly shaped features in different organisms need not have the same biological basis.

Focal contacts and chromatin loops

Some pairs of regions produce localized contact enrichments, often called peaks or loops. Many mammalian loop anchors contain convergently oriented CTCF-binding sites.

The loop extrusion model explains important features of these patterns: cohesin can enlarge a DNA loop, while appropriately oriented bound CTCF can act as a barrier. Hi-C patterns, together with imaging and perturbation experiments, support this mechanism for many mammalian loops and domains.

Not all contact enrichment is produced by loop extrusion. Compartments, transcription-associated interactions, and other chromatin processes also contribute.

Hi-C can detect contacts involving regulatory elements and promoters, but proximity alone does not establish that one element regulates a particular gene.

Development and related methods

Chromosome conformation capture was introduced in 2002 as a targeted method for studying spatial associations between genomic loci. Subsequent techniques expanded the number of interactions that could be examined.

Hi-C was introduced in 2009 by Erez Lieberman-Aiden and colleagues, providing a genome-wide approach that helped establish large-scale compartment organization. In situ Hi-C, described in 2014 by Suhas Rao and colleagues, improved the recovery of informative contacts and enabled substantially finer mapping of chromatin architecture.

Several related approaches address different experimental needs:

  • Capture Hi-C enriches selected sequences from a Hi-C library, increasing coverage around chosen regions without sequencing the entire library equally deeply.
  • Single-cell Hi-C assigns contacts to individual cells or nuclei. It reveals cell-to-cell variation, but individual contact maps are generally sparse.
  • Micro-C commonly uses micrococcal nuclease to fragment chromatin near the nucleosome scale, enabling finer analysis of short-range folding.
  • HiChIP and related protein-enriched assays preferentially recover contacts associated with a selected protein or chromatin modification. Their maps are targeted by biochemical enrichment and are not directly equivalent to an unbiased genome-wide contact survey.

Different protocols trade off input requirements, coverage, specificity, cost, and achievable resolution. Maps from different assay types are not automatically interchangeable.

Applications

Genome organization and regulation

Hi-C is used to compare chromosome architecture across cell types, developmental stages, species, and experimental perturbations. Combined with gene-expression, chromatin-accessibility, or protein-binding measurements, it helps characterize relationships between genome folding and regulation.

Changes in contact patterns do not by themselves establish causation. Structural changes may influence transcription, result from altered activity, or accompany another process affecting both.

Genome assembly

Hi-C also supports genome assembly. Because sequences from the same chromosome generally contact one another more frequently than sequences from different chromosomes, contact data can help assign assembled fragments to chromosomes and determine their order and orientation.

Assembly scaffolding must account for repetitive DNA, heterozygosity, rearrangements, and unusual biological contact patterns. An assembly constructed using Hi-C is not independently validated merely because its resulting contact map appears orderly.

Structural variation and comparative genomics

Large genomic rearrangements can create characteristic discontinuities or unexpected blocks in contact maps. Hi-C can therefore contribute to research on chromosome rearrangements and genome evolution.

Interpreting these patterns requires attention to copy-number variation, reference-genome differences, mapping artifacts, and sample heterogeneity. Apparent contact changes can arise from altered genome sequence or chromosome content rather than altered folding alone.

Limitations and interpretation

Bulk measurements average across cells. Conventional Hi-C combines contacts from many cells, often encompassing different cell-cycle stages and biological states. A strong feature in the aggregate map need not represent a structure present continuously in every cell.

Contact frequency is not a direct distance measurement. Crosslinking, accessibility, fragmentation, and ligation all affect capture probability. Higher contact frequency often corresponds to closer spatial association, but no universal conversion turns a contact count into a distance in nanometres.

Pairwise maps do not resolve simultaneous multiway organization. Standard analysis records pairwise associations. It generally cannot determine whether several contacts coexist in one cell or occur separately across different cells.

Fine-scale resolution is data-intensive. Smaller bins create many more potential bin pairs, most of which may have few or no observations. Sequencing depth, library complexity, genome size, and the feature being studied all affect effective resolution.

Some regions are difficult to measure. Repetitive sequences, poorly assembled regions, and closely related duplicated sequences may be hard to map uniquely. These difficulties can obscure biologically important chromosome regions.

Computational calls are scale- and method-dependent. Domain boundaries, loops, and compartment assignments can change with filtering, normalization, bin size, and detection thresholds. Reproducibility and sensitivity analysis are therefore important when interpreting differences.

A contact map does not specify a unique three-dimensional structure. Computational models can identify spatial arrangements consistent with selected aspects of the data, but multiple conformations may explain the same population-averaged map. Orthogonal methods, including fluorescence in situ hybridization, provide complementary measurements of distances, variability, and nuclear position.